deleting a netbox makes data-collection stop

[jmbredal]

When deleting a netbox the table ipdevpoll_job_log seems to be locked for updates. As every ipdevpoll job wants to update this table they all is kept waiting when the delete takes a long time.

Related reference in local ticket system:
https://rt.uninett.no/Ticket/Display.html?id=115204

---

Imported from Launchpad using lp2gh.

• date created: 2016-08-16T07:57:23Z
• owner: john-m-bredal
• the launchpad url was https://bugs.launchpad.net/bugs/1613590

[lunkwill42]

Observing this again now. It seems that what may actually be happening here is that while PostgreSQL is running cascading deletes, ipdevpoll still doesn't know that the netbox in question is being deleted, so the jobs related to this device are still running. Any attempt to insert an ipdevpoll_job_log entry related to this netbox will be blocked by PostgreSQL, eventually causing all available database connections in an ipdevpoll process to be blocked.

One suggested solution would be to introduce a deletion flag on netbox, which would cause ipdevpoll to stop working on it. This overlaps with one of the discussed solutions for #1313.